Filtration Physiology Sits Inside the Organ Clock, Not Outside It

Question: How do organ-specific physiological confounders such as renal filtration and hepatic clearance bias proteomic and metabolomic biological age clocks, and what statistical calibration approaches have been used to remove them?

Answer

Taken across these papers, the position this survey takes is that organ-specific physiological confounders are not an external nuisance attached to plasma and serum omic clocks — they are a substantial fraction of what the clocks measure. Filtration-handled proteins rank among the strongest "aging" analytes in the largest cohorts [3], impaired kidney function is documented as a driver of transitions into higher proteomic age acceleration [7], and eGFR and albuminuria were reported as the dominant determinants of an inflammation-related serum proteome [9]. Against that, the retrieved calibration literature is remarkably thin: exactly one named procedure exists — Filtrate-Aware Calibration, developed for a urinary peptidomic clock and conditioning on eGFR and UACR [4] — alongside implicit residualization [9], derivation-cohort restriction [4], and matrix substitution [11]. Critically, the age-prediction methodology literature indicates that calibration of this kind cannot be validated by better chronological-age prediction, because age-supervised training preferentially weights the stable, shared variance that organ endowment and clearance capacity themselves generate [19] [21].

Filtration proxies are among the strongest circulating "aging" signals, so renal physiology is inside the clock rather than outside it

In the Lothian Birth Cohort 1936, the plasma proteins most strongly associated with all-cause mortality were GDF15 (HR per SD 1.53), CST3 (1.48) and COL18A1 (1.47), with the 202 mortality-associated proteins enriched for liver and immune-related biological processes [3]. CST3 is cystatin C, a standard filtration marker, which means a glomerular-filtration readout was among the top "proteomic aging" analytes in that cohort. In the KORA longitudinal proteomic clock (9,105 samples, 4,100 individuals), dyslipidemia and impaired kidney function tracked transitions to increases in age acceleration within individuals, and the implicated pathways included liver-derived plasma proteins as key contributors [7]. In 683 children with CKD profiled on an inflammation panel, eGFR and albuminuria were the dominant determinants of the inflammation-related serum proteome [9]. Even in a healthy pediatric population, creatinine rose monotonically with age across infant, pre-teen and post-pubertal strata [5], so an age-trained model on any panel containing creatinine learns renal maturation as aging.

Only one clock makes organ-specific physiological calibration explicit, and it is renal only

The urinary peptidomic clock was developed as an unconditioned model (UPBioAge) in a kidney-function-preserved derivation cohort (n = 1,811) and then conditioned on eGFR and UACR by Filtrate-Aware Calibration fitted in an independent kidney-diverse cohort (n = 7,798), producing k-UPBioAge [4]. After standard age-bias correction, the conditioned clock estimated chronological age with a calibrated holdout MAE of 4.91 years and r = 0.945 [4]. Each SD of kidney-conditioned age acceleration was associated with all-cause mortality (HR 1.48), incident coronary artery disease (1.44), heart failure (1.27) and CKD progression (1.35), and no comorbidity interaction survived multiple-testing correction [4]. Critically, no association was evident at eGFR 15–29 mL/min/1.73 m², indicating the calibration does not rescue the extreme-filtration stratum [4]. No retrieved paper reports an analogous calibration for hepatic clearance, despite liver-derived proteins being flagged as dominant contributors elsewhere [7] [3].

The substitutes actually used — residualization, restriction, covariate adjustment — are ad hoc and each carries a documented failure mode

Residualization against filtration markers is used and works well enough to be reported: a four-protein inflammation score (CD40, CD137, PD-L1, CX3CL1) was defined from protein residuals independent of eGFR and albuminuria and predicted CKD progression beyond established risk factors (adjusted HR 1.11), replicating in UK Biobank [9]. Restriction to a preserved-function derivation cohort is used as a pre-calibration device but is explicitly insufficient on its own, since FAC still had to be fitted in a separate kidney-diverse cohort before conditioning could be applied [4]. Covariate adjustment in survival models is the commonest fallback and also the weakest, since the clock's association survived eGFR adjustment in Cox models yet vanished in advanced CKD [4]. The adjustment covariate itself is unstable: race-neutral eGFR equations diverge enough that age-adjusted CKD prevalence ranged from 8.5% to 10.8% depending on the equation chosen across nearly two million individuals [14], and creatinine-based and cystatin C-based eGFR gave discordant conclusions in interventional and observational settings, with the creatinine signal not corroborated by the muscle-mass-independent marker [15] [16]. Creatinine-based equations show systematically weaker associations than cystatin C-based ones, attributed to muscle-mass confounding [24]. One practical enabler is that eGFR can be recovered retrospectively from long-term frozen serum and plasma with strong agreement to point-of-care whole blood (weighted kappa 0.90–0.95), so calibration is feasible even in cohorts that never measured renal function at draw [6].

Calibrating out organ physiology is in direct tension with what makes clocks prognostic

Models with lower chronological age prediction accuracy — simpler, more heavily regularized ones — showed higher sensitivity to disease-relevant change than more accurate counterparts, because optimizing age accuracy drives models toward low-variance shared aging features and away from higher-variance signals informative about health [18]. Age-supervised brain models demonstrably downweight the individual-trajectory features an aging-rate biomarker must capture, and flagged "accelerated aging" in low-birth-weight participants with no longitudinal evidence of it, converting stable anatomical differences into apparent acceleration [19]. Formally, squared-error age loss penalizes within-age output dispersion without defining its biological direction, and exact age recovery eliminates the age gap even when heterogeneity remains in the measurements [21]. Composite ageing measures exhibit feature convergence that structurally constrains cross-sectional stratification and longitudinal monitoring, and the same limitation persists in molecular implementations [17]. Empirically, chronological-age-trained epigenetic clocks produced sporadic, non-replicable changes that disappeared when retrained for test-retest reliability, whereas mortality- or pace-trained clocks were consistent [20]. If nephron endowment or hepatic clearance capacity is a stable personal attribute, calibrating it away and preserving it are both defensible and lead to opposite biological claims.

Beyond the blood: matrix, timing and tissue reference choices generate their own organ gradients

Switching matrix changes the confounder profile rather than eliminating it: urine proteomics on Olink Explore 3072 found 157 age-associated proteins with tissue-specific expression in hepatocytes, immune cells and proximal tubule cells, and detectability itself was bimodal, with 47% of proteins detectable in under 20% of samples [12]. Urinary vesicle MUC1 was proposed precisely because blood markers are confounded by muscle mass and urine "bypasses systemic metabolic noise" [11]. Sample timing is an under-adjusted physiological confounder: 84 of 853 serum metabolites were associated with blood draw time within a short daytime window, with measurable impact on false and missed discovery and on effect-size estimates [13]. In kidney tissue, adjusting for donor age and procurement altered pathway-level results, and aging-associated processes were enriched in older reference tissues in a way that overlaps disease mechanisms [8]. Pharmacological physiology behaves similarly: medication initiation altered organ clocks through specific drug-targeted proteins such as renin and APOB rather than through generalized organ aging [1], and platform discordance between affinity proteomics methods remains substantial for low-abundance, high-proteoform-diversity proteins [23].

Conflicts and gaps

Two disagreements are unresolved. First, whether physiological calibration improves or destroys a clock: the disease-detection literature argues accuracy-optimized models are less informative [18], while the calibration literature treats better age prediction as the criterion of success [4] [22]. Second, whether organ-specific clocks index organ biology or shared fibrosis and inflammation: a 50-peptide COL1 degradation signature achieved AUC 0.935 in an external cohort and performed comparably across liver disease (0.917), CKD (0.951) and heart failure (0.950) [10], which undermines the premise that a "liver clock" is liver-specific. Hepatic clearance has no calibration equivalent at all, and where hepatobiliary metabolites were examined in extreme old age, higher bile acids and bilirubin were associated with lower mortality [2] — the opposite direction to a simple clearance-deficit model. No retrieved study reports a clock calibrated jointly on eGFR, UACR and a hepatic clearance measure, and none reports calibration against measured (rather than estimated) GFR.

What would change the answer

A clock trained and validated with simultaneous measured GFR (iohexol or iothalamate clearance) and measured hepatic clearance as conditioning covariates, in a cohort spanning preserved to severely impaired function, would directly test whether Filtrate-Aware Calibration generalizes beyond eGFR/UACR and transfers to the liver axis. A within-person design exploiting pre- versus post-transplant or pre- versus post-initiation-of-dialysis samples would separate clearance effects from tissue aging, since the same organ is observed at two filtration states. Negative-control outcomes — a clock predicting dialysis vintage better than comorbidity burden — would quantify residual filtration loading. Finally, comparing a calibrated clock against an uncalibrated one on incident disease rather than on chronological-age MAE, in the spirit of the disease-sensitivity benchmarking already applied to brain-age models [18] and to proteomic organ ages against GrimAge2 and neuroimaging [3], would settle whether removing filtration variance discards confounding or discards signal.

References

[1] Neirynck, R. E., Chirinos, J. A., Van Damme, M., Coussement, L., Segers, P., De Buyzere, M., Rietzschel, E. R., De Meyer, T.. "Longitudinal dynamics of organ-specific proteomic aging clocks over a decade of midlife". bioRxiv. 2026-02-18T00:00:00.000Z. URL: https://doi.org/10.64898/2026.02.17.706320

[2] Monti, S., Lustgarten, M. S., Huang, Z., Song, Z., Ellis, D. S., Tian, Q., Ferrucci, L., Rappaport, N., Andersen, S., Perls, T., Sebastiani, P.. "METABOLOMIC SIGNATURES OF EXTREME OLD AGE: FINDINGS FROM THE NEW ENGLAND CENTENARIAN STUDY". bioRxiv. 2025-09-16T00:00:00.000Z. URL: https://doi.org/10.1101/2025.09.10.675341

[3] Pyrgioti, M., Eguiagaray, I. M., Redmond, P., Corley, J., Bastin, M. E., Valdes Hernandez, M., Russ, T. C., Wardlaw, J. M., Hannon, E., Deary, I. J., Walker, K. A., Tucker-Drob, E. M., Cox, S. R., Marioni, R. E., Harris, S. E.. "Multimodal Ageing Biomarkers and Plasma Proteomic Signatures Associated with All-Cause Mortality". medRxiv. 2026-03-10T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.09.26347914

[4] Biglari, S., Jaimes-Campos, M. A., Siwy, J., Latosinska, A., Mischak, H., Nawrot, T. S., Staessen, J. A., Martens, D. S., Banasik, M.. "A kidney-conditioned urinary peptidomic biological ageing clock predicts all-cause mortality and age-related health outcomes". medRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.01.26361640

[5] Thachil, A., Mandal, R., Wishart, D. S., Blydt-Hansen, T.. "Age-related differences in metabolome composition in a healthy pediatric population: A pilot analysis". bioRxiv. 2026-09-17T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.13.750246

[6] French, S. R., Culwell, G. C., Wiskoski, H. E., Arias, J. C., Zahra, S., Escareno, C. E., Heitkamp, E. N., Garcia, A. R., Vidana, P., Quijada, J. B., Kasparov, R., Vitali, F., Bedrick, E. J., Mushtaq, R., Alexander, G. E., Weinkauf, C. C.. "Reliable quantification of renal function from frozen blood samples". medRxiv. 2026-06-22T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.12.26355531

[7] Lermer, R., Qian, X., Pusch, S. T., Schaefer, T., Adam, J., Li, L., von Scheidt, M., Linkohr, B., Peters, A., Mann, M., Schunkert, H., Heinig, M.. "Intraindividual longitudinal changes of the proteome reveal dynamic markers of biological aging". bioRxiv. 2026-10-06T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.01.755909

[8] Menon, R., Kimmel, P. L., Otto, E. A., Subramanian, L., OConnor, C. L., Godfrey, B., Smith, C., Alakwaa, F. M., Berthier, C. C., Sarwal, M., Woodle, E. S., Pyle, L., Choi, Y. J., Ladd, P., Sedor, J. R., Rosas, S. E., Waikar, S. S., Naik, A. S., Melo Ferreira, R., Eadon, M. T., Bitzer, M., Bjornstad, P., Hodgin, J. B., Kretzler, M.. "Not all controls are made equal: Definition of human kidney reference samples by single cell gene expression profiles". medRxiv. 2026-01-16T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.17.25324134

[9] Bartolomaeus, H., Reitmeir, R., Versnjak, J., Hofstetter, J., Behrens, F., Yarritu, A., Bonnekoh, P., Liebau, M. C., Bayazit, A. K., Duzova, A., Canpolat, N., Kaplan Bulut, I., Azukaitis, K., Obrycki, L., Wilck, N., Weitz, M., Zernecke, A., Melk, A., Querfeld, U., Kelm, M., 4C Study Consortium,, Schaefer, F., Holle, J.. "Inflammation-based risk score predicts kidney survival in children and adults with chronic kidney disease". medRxiv. 2026-07-02T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.01.26357013

[10] Mina, I. K., Hussain, Y., Siwy, J., Catanese, L., Rupprecht, H., Beige, J., Staessen, J. A., Metzger, J., Persson, F., Rossing, P., Delles, C., Schanstra, J. P., Bannaga, A., Vlahou, A., Mischak, H., Arasaradnam, R. P., Latosinska, A.. "Urinary collagen type I degradation products as common fibrosis biomarkers in chronic diseases". medRxiv. 2026-08-31T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.26.26361420

[11] Nishimura, T., Harita, Y., Hirakawa, Y., Takizawa, K., Fujishiro, J., Ogawa, S., Kajiho, Y., Kanda, S., Kushima, R., Omori, T., Hamasaki, Y., Gotoh, Y., Miura, K., Fujita, N., Okamoto, T., Hisano, M., Nangaku, M., Kato, M.. "A non-invasive liquid biopsy resolves the diagnostic blind spot in chronic kidney disease". medRxiv. 2026-06-17T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.09.26355142

[12] Meier, M., Schoberth, M., Li, Y., Borisov, O., Uluvar, B., Willecke, C., Meiselbach, H., Eckardt, K.-U., Schilling, O., Kottgen, A., Haug, S.. "Characterization of the Urine Proteome Using the Olink Explore 3072 Platform". medRxiv. 2025-12-09T00:00:00.000Z. URL: https://doi.org/10.64898/2025.12.08.25341840

[13] Nagarajan, P., Faquih, T., Yu, B., Dashti, H. S., Noordam, R., Xiao, Q., Kaplan, R., Qi, Q., Isasi, C. R., Saxena, R., Scheer, F. A. J. L., Redline, S., Sofer, T., Wang, H.. "The Diurnal Rhythmicity of Untargeted Metabolomic Profiles in Humans". medRxiv. 2026-04-29T00:00:00.000Z. URL: https://doi.org/10.64898/2026.04.28.26347027

[14] James A. Diao, Morgan Sanchez, Adir Sommer, Keren Cohen, Maya Makov-Assif, Marinka Zitnik, Ben Reis, Ran D. Balicer, Noa Dagan, Arjun K. Manrai. "Kidney function and kidney failure prediction in a large multiethnic population". arxiv. 2026-07-27T00:00:00.000Z. URL: https://arxiv.org/abs/2607.22504

[15] Rong, F., Wu, Z., Xu, Y., Liu, W., Zhou, G., Ding, W., Cao, J., Xiao, G., Xu, D., Zhou, H.. "Effects of Ergothioneine Supplementation on Glomerular Filtration and Patient-Reported Urological Symptoms in Adults with Early Renal Function Decline: A Single-Center, Open-Label, Self-Controlled Trial". medRxiv. 2026-07-09T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.08.26356008

[16] Irifuku, T., Kashiwado, S., Masaki, T.. "Switching from febuxostat to dotinurad in patients with chronic kidney disease and hyperuricemia: a single-center, non-randomized study". medRxiv. 2026-07-18T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.16.26358294

[17] Hu, I.. "House of Clocks: On the Misuse of Ageing Composite Measures". bioRxiv. 2025-05-28T00:00:00.000Z. URL: https://doi.org/10.1101/2025.05.24.655934

[18] Schulz, M.-A., Siegel, N. T., Ritter, K.. "Brain-age models with lower age prediction accuracy have higher sensitivity for disease detection". bioRxiv. 2025-10-21T00:00:00.000Z. URL: https://doi.org/10.1101/2024.03.28.587212

[19] Grodem, E. O. S., Smith, S. M., Vidal-Pineiro, D., Elliott, M. L., for the Alzheimer's Disease Neuroimaging Initiative,, Walhovd, K. B., Fjell, A. M.. "Reconsidering Brain Age: Why Age-Prediction Models Fail as Measures of Brain Aging". medRxiv. 2026-05-08T00:00:00.000Z. URL: https://doi.org/10.64898/2026.05.07.26352620

[20] Borrus, D. S., Sehgal, R., Armstrong, J. F., Kasamoto, J., Gonzalez, J., Higgins-Chen, A. T.. "When to Trust Epigenetic Clocks: Avoiding False Positives in Aging Interventions.". bioRxiv. 2024-10-25T00:00:00.000Z. URL: https://doi.org/10.1101/2024.10.22.619720

[21] Song, S., Liang, Q., Gao, Q.. "Inferential boundaries of age prediction: why prediction does not establish biological age measurement". bioRxiv. 2026-09-29T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.23.753690

[22] Flores, M., Pellegrini, M.. "Improving the Accuracy of Forensic Age Estimation Through Bias Reduction". bioRxiv. 2026-06-03T00:00:00.000Z. URL: https://doi.org/10.64898/2026.05.30.728628

[23] Zuehlke, B. M., Schwecke, T., Weizmann Shapira, T., Niewienda, A., Yang, A., Amari, F., Textoris-Taube, K., Sinn, L. R., Lehmann, K., Gille, C., Vingron, M., Mülleder, M., Ralser, M.. "A draft atlas of the human plasma proteome at isoform resolution". bioRxiv. 2026-09-28T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.26.754256

[24] Komleva, Y., Koenig, M., Ittermann, T., Friedrich, N., Petersmann, A., Nauck, M., Voelzke, H., Mannaa, M., Gollasch, M.. "Kidney Function Estimation and its Relationship with Frailty in Older Adults: Insights from SHIP-TREND". medRxiv. 2025-10-30T00:00:00.000Z. URL: https://doi.org/10.1101/2025.10.28.25339027

How this survey was made

This survey answers one question: “How do organ-specific physiological confounders (e.g., renal filtration, hepatic clearance) bias proteomic and metabolomic biological age clocks, and what statistical calibration approaches have been used to remove them?”

We keep our own collection of recent preprints and refresh it every day from bioRxiv, medRxiv, arXiv, PsyArXiv, SocArXiv and ChemRxiv. For each survey, we search that collection for recent work and keep the strongest matches: papers that use the same terms as the question, and papers on the same topic in different words.

The text was written from the titles and abstracts of those papers; the full texts were not read. Statements that draw on a particular paper are marked with a number in brackets. The reference list just above is the 24 papers whose work this text draws on, chosen from the 117 papers those searches returned; the rest of the shortlist is not printed. It is not a complete review of everything published on the question.

Treat this as a starting point and a reading list, not a peer-reviewed review. The searches behind this survey ran on October 22, 2026 and covered work from the previous two years.